The present invention provides a dialogue system in which semantic ambiguity is reduced by selectively choosing which semantic structures are to be made available for parsing based on previous information obtained from the user or other context information. In one embodiment, the semantic grammar used by the parser is altered so that the grammar is focused based on information about the user or the dialogue state. In other embodiments, the semantic parsing is focused on certain parse structures by giving preference to structures that the dialogue system has marked as being expected.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A method of utilizing and modifying a grammar in a semantic parser, the method comprising: identifying a set of surface semantic non-terminals from user input from a first user using a context-free grammar; using a language-independent semantic description language grammar that is separate from the context-free grammar and that describes relationships between semantic non-terminals to identify a semantic parse based in part on the identified semantic non-terminals; and changing the semantic description language grammar based on a change in users from the first user to a second user.
2. The method of claim 1 wherein changing the semantic description language grammar comprises deactivating a semantic entity in the grammar such that the semantic entity remains in the grammar but is not used to identify semantic parses.
3. The method of claim 1 wherein changing the semantic description language grammar comprises activating a semantic entity in the grammar that has previously been deactivated.
4. The method of claim 1 wherein changing the semantic description language grammar comprises combining a second semantic description language grammar with the semantic description language grammar.
5. The method of claim 1 wherein changing the semantic description language grammar comprises inserting a semantic entity into the semantic description language grammar.
6. The method of claim 1 wherein changing the semantic description language grammar comprises deleting a semantic entity from the semantic description language grammar.
7. The method of claim 6 wherein deleting a semantic entity comprises deleting a non-terminal from the semantic description language grammar.
8. A method of parsing text to form a semantic structure, the method comprising: setting a focus structure based on a dialogue on a state determined after a first input wherein the focus structure indicates an expected semantic structure and is separate from grammars used to identify possible semantic structures; identifying possible semantic structures for text associated with a second input using the grammars; and returning only those semantic structures that correlate to the focus structure.
9. The method of claim 8 wherein returning only those semantic structures that correlate to the focus structure comprises only returning those semantic structures that completely define a non-terminal in the focus structure.
10. The method of claim 9 wherein returning only those semantic structures that correlate to the focus structure further comprises only returning those semantic structures that span all of the text.
11. A computer-readable storage medium having computer-executable instructions for performing steps comprising: setting a focus semantic parse that is separate from grammars used to form candidate semantic parses and that is based on a dialog state determined after a first user input; parsing a text associated with a second user input to form candidate semantic parses using the grammars; comparing the candidate semantic parses to the focus parse; and giving preference to a candidate semantic parse that best matches the focus parse.
12. The computer-readable storage medium of claim 11 wherein giving preference to a candidate semantic parse comprises returning only the candidate semantic parse that best matches the focus parse.
13. The computer-readable storage medium of claim 11 wherein giving preference to a candidate semantic parse comprises giving preference to a candidate semantic parse that fully defines a non-terminal in the focus parse.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 29, 2005
June 30, 2009
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.